Misclassification of Safety Net Hospitals With National Data

Abstract

We compliment the recent retrospective study in CHEST (July 2021) by Williamson et al.1Williamson C.G. Hadaya J. Mandelbaum A. et al.Outcomes and resource use associated with acute respiratory failure in safety net hospitals across the United States.Chest. 2021; 160: 165-174Abstract Full Text Full Text PDF PubMed Scopus (5) Google Scholar Using the Healthcare Cost and Utilization Project’s (HCUP) National Inpatient Sample (NIS), the authors investigated the association of safety net hospital status (SNH) on outcomes of acute respiratory failure. SNH status was based on the percentage of patients in each hospital whose primary insurance was either Medicaid or who were uninsured, a method loosely based on prior HCUP methods.2Sutton J.P. Washington R.E. Fingar K.R. Elixhauser A. Stastical Brief #213: Characteristics of Safety-Net Hospitals, 2014.https://www.hcup-us.ahrq.gov/reports/statbriefs/sb213-Safety-Net-Hospitals-2014.jspGoogle Scholar,3Reiter K.L. Jiang H.J. Wang J. Facing the recession: how did safety-net hospitals fare financially compared with their peers?.Health Serv Res. 2014; 49: 1747-1766PubMed Google Scholar The authors defined high-burden SNHs as those in the top tertile (>66th percentile) and low-burden SNHs as those in the bottom tertile (<33rd percentile) of percentage of patients with Medicaid or who were uninsured. The authors observed that admission to a high-burden SNHs was associated with higher odds of death and odds of tracheostomy. The authors’ findings are significant, especially given inequities and disparities identified during the COVID-19 pandemic. However, their approach in defining the exposure variable, SNH, deviates from past standards published by HCUP and likely results in misclassification of some hospitals.2Sutton J.P. Washington R.E. Fingar K.R. Elixhauser A. Stastical Brief #213: Characteristics of Safety-Net Hospitals, 2014.https://www.hcup-us.ahrq.gov/reports/statbriefs/sb213-Safety-Net-Hospitals-2014.jspGoogle Scholar Previous methodologic evaluations of SNHs by Sutton et al2Sutton J.P. Washington R.E. Fingar K.R. Elixhauser A. Stastical Brief #213: Characteristics of Safety-Net Hospitals, 2014.https://www.hcup-us.ahrq.gov/reports/statbriefs/sb213-Safety-Net-Hospitals-2014.jspGoogle Scholar and Reiter et al3Reiter K.L. Jiang H.J. Wang J. Facing the recession: how did safety-net hospitals fare financially compared with their peers?.Health Serv Res. 2014; 49: 1747-1766PubMed Google Scholar defined SNHs as hospitals in the top quartile for percentage of Medicaid and uninsured patients within a given state. Using state-specific cutoffs for the top quartile is critical due to the wide variability in percentage of Medicaid and uninsured patients from state to state. The NIS does not indicate in which state a given hospital is located but does include hospital region (Northeast, Midwest, South, and West). Using the 2018 NIS, we determined the 33rd and 66th percentiles of Medicaid and uninsured patients for all hospitals and then specifically for each region (568 hospitals had no region listed) (Table 1). We further determined the number of hospitals that were misidentified by using national cutoffs. Overall, 350 hospitals (7.7% of all hospitals in the 2018 NIS) were misidentified: 42 in the Northeast, 133 in the Midwest, 46 in the South, and 129 in the West.Table 1Misclassification of Safety Net Hospitals Using Only National-Level DataRegionPercentage of Medicaid and Uninsured Patients, %No. of HospitalsTertile 1 (Low Burden)aWilliamson et al1 use tertiles (ie, <33rd percentile, 33rd to 66th percentile, and >66th percentile) to define low- and high-burden safety net hospitals.Tertile 2aWilliamson et al1 use tertiles (ie, <33rd percentile, 33rd to 66th percentile, and >66th percentile) to define low- and high-burden safety net hospitals.Tertile 3 (High Burden)aWilliamson et al1 use tertiles (ie, <33rd percentile, 33rd to 66th percentile, and >66th percentile) to define low- and high-burden safety net hospitals.Misidentified as Safety Net Hospital StatusFailed to Be Identified as Safety Net Hospital StatusNational<16.216.2-27.2>27.2……Northeast<15.415.4-24.9>24.9042Midwest<14.014.0-23.5>23.50133South<16.916.9-28.2>28.2460West<21.221.2-33.5>33.51290a Williamson et al1Williamson C.G. Hadaya J. Mandelbaum A. et al.Outcomes and resource use associated with acute respiratory failure in safety net hospitals across the United States.Chest. 2021; 160: 165-174Abstract Full Text Full Text PDF PubMed Scopus (5) Google Scholar use tertiles (ie, <33rd percentile, 33rd to 66th percentile, and >66th percentile) to define low- and high-burden safety net hospitals. Open table in a new tab The degree of misidentification would vary from year to year, and we provide data for only 2018. However, failure to account for regional variation in cutoffs for SNHs could lead to erroneous conclusions. Adjustment regression models for hospital region would be insufficient to address the issue of defining the primary exposure. We believe further exploration with region- or state-specific definitions for SNHs are necessary to understand the true scope of differential outcomes and disparities. Outcomes and Resource Use Associated With Acute Respiratory Failure in Safety Net Hospitals Across the United StatesCHESTVol. 160Issue 1PreviewAfter accounting for differences between patient cohorts, high safety net burden was associated independently with inferior clinical outcomes and increased costs after ARF hospitalizations. These findings emphasize the need for health care reform to ameliorate disparities within these safety net centers, which treat our most vulnerable populations. Full-Text PDF ResponseCHESTVol. 160Issue 4Preview“Facts are stubborn things, but statistics are pliable.”-Mark Twain Full-Text PDF